Junify
Managing shared accounts

Do passwords still need to be changed every 90 days?

Not if only one person knows it. NIST SP 800-63B-4 bans forced scheduled changes and requires one after a compromise. Shared passwords change when people leave.

Updated October 16, 2026

Whether you keep changing passwords on a schedule depends on who knows the password. NIST SP 800-63B-4 bans scheduled changes, and the guideline assumes only the user knows the password. PCI DSS drops them where a second factor is used. A shared account used by several people breaks NIST’s assumption.

What NIST SP 800-63B-4 says

NIST SP 800-63B-4 is NIST’s guideline for how services check who is signing in. NIST published the final version in July 2025. Password rules are in section 3.1.1.2. They’re addressed to verifiers, the services that check the password, and CSPs, the organizations that issue and manage the accounts.

Topic What section 3.1.1.2 says
Scheduled changes “Verifiers and CSPs SHALL NOT require subscribers to change passwords periodically”
Change after a compromise They shall force a change when there’s evidence the password has been compromised
Length At least 15 characters when the password is the only factor, and at least 8 when it’s used with a second factor
Character rules They shall not impose composition rules, such as requiring a mix of character types
Common and leaked passwords They shall check new passwords against a blocklist of common, expected or compromised passwords

Keep the audience in mind when you cite this section. It tells services not to force scheduled changes on their users. If your company drops scheduled changes from its employee policy because of it, you’re applying NIST’s reasoning to your own staff, which the section itself doesn’t address.

The guideline’s introduction says users are responsible for protecting their sign-in secrets and “not disclosing them to others.” Read together, the rule against scheduled changes appears to assume that only the user knows the password.

The earlier version, published in 2017, said services “SHOULD NOT” require changes for no reason, such as on a schedule. Revision 4 made that a “SHALL NOT.”

What PCI DSS says

PCI DSS is the data security standard for payment card data, set by the PCI Security Standards Council (PCI SSC). Version 4.0.1, published in June 2024, replaced version 4.0, which PCI SSC retired on December 31, 2024. PCI SSC says 4.0.1 added and removed no requirements.

Requirement 8.3.9 covers passwords that are the only factor for user access, meaning any setup with no second factor. In that case, one of two things must be true. Either passwords change no less often than every 90 days, or the system analyzes each account’s security in real time and decides access from that. PCI SSC’s FAQ 1590 says 8.3.9 doesn’t apply to systems where multi-factor authentication is used.

PCI DSS also wants a change when something goes wrong. Requirement 8.3.8 asks for written password policies that tell users to change a password if they suspect or know it’s been compromised, and how to report it.

Source Scheduled changes When a change is required
NIST SP 800-63B-4 (section 3.1.1.2) Shall not be required When there’s evidence of compromise
PCI DSS v4.0.1 (requirements 8.3.8 and 8.3.9) Every 90 days only when the password is the only factor, unless access is decided by real-time risk analysis Policy must tell users to change it on suspicion or knowledge of compromise

Why scheduled changes can go

A scheduled change limits how long a stolen password keeps working when nobody notices the theft. Change every 90 days, and an unnoticed leak stops working within 90 days at most. That only helps if you have no better way to notice.

Both sources expect a change when there’s evidence of a compromise. Signs of a compromise include a breach notice from the service, a sign-in alert, a match on a breach-check site, or an odd entry in the account’s sign-in history. Once you change the password at that point, it stops working sooner than if you’d waited for the next deadline. Until then, the defense is a long, unique password checked against a blocklist, plus a second factor.

Why shared accounts don’t fit

Dropping scheduled changes rests on two conditions. Only the owner knows the password, and anyone else’s use leaves evidence. A shared account meets neither.

First, several people know the password from day one. NIST assumes people don’t tell others their password, and a shared account works only because they do. Second, the group changes. When someone leaves or moves to another team, they still remember the password. For a personal account, that would count as a stranger knowing it. For a shared account, there’s no evidence of it. The service’s records show only the account name, so a former user’s sign-in looks like anyone else’s.

So on a shared account you can’t switch from “change on a schedule” to “change on evidence.” The change has to happen when the group that knows the password changes: a departure, a role change, the end of a contractor’s work. NIST SP 800-53 Rev. 5, the control catalog federal systems must use, covers this in control AC-2 k. It asks for a process to change a shared or group account’s password when someone is removed from the group.

The new password then goes to everyone who stays, so the next departure means doing it all again. Dropping scheduled changes doesn’t remove this work.

When you can drop scheduled changes

Condition Personal account Shared account
Only the owner knows the password Yes No, unless people can sign in without being shown it
You can spot a compromise Sign-in alerts, breach checks, sign-in history A former user’s sign-in looks like everyone else’s
Long password that isn’t reused 15 characters or more when it’s the only factor, checked against a blocklist Same
Second factor Turn it on where the service offers it A code on one person’s phone blocks everyone else when they’re out
Cutting off one person Disable their account Only by changing the password and sending it out again

Until a shared account meets the conditions in the first and last rows, you have no basis for dropping its scheduled changes. Managing shared accounts compares the ways to meet those conditions. What is two-step verification? covers sharing a second factor.

Frequently asked questions

Can we drop the 90-day rule from our password policy?

For accounts only one person uses, NIST's approach is to replace it with a change when there's evidence of compromise, plus long passwords checked against a blocklist. Keep a separate rule for shared accounts.

Does PCI DSS still require changing passwords every 90 days?

Only where the password is the only sign-in factor (requirement 8.3.9), and even then you can analyze account risk in real time instead. PCI SSC says 8.3.9 doesn't apply where multi-factor authentication is used.

How often should a shared account's password change?

Not on a set number of days. Change it whenever the group that knows it changes, such as a departure, a role change or the end of a contract. To avoid that, let people sign in without being shown the password.

What counts as evidence that a password was compromised?

Anything that suggests someone else used it. Examples are a breach notice from the service, a sign-in alert you don't recognize, a match on a breach-check site, or an odd sign-in in the account's history.

How Junify handles this

Doing this by hand
To keep shared accounts safe, you track who knows each password, and change it and send out the new one whenever someone leaves. You also keep a separate record of who did what under the same login.
With Junify
With Junify, change the shared account's password when your admin adds it. People then verify on their own phone and click to sign in without seeing it. So a departure or role change doesn't mean changing it and sending the new one to everyone who stays. The admin removes that person's access in Junify, and each use is recorded by person.

Sources

More on this topic